diff --git a/milli/src/update/clear_documents.rs b/milli/src/update/clear_documents.rs
index 250e4b13a..6d7dd72b8 100644
--- a/milli/src/update/clear_documents.rs
+++ b/milli/src/update/clear_documents.rs
@@ -56,6 +56,7 @@ impl<'t, 'u, 'i> ClearDocuments<'t, 'u, 'i> {
         docid_word_positions.clear(self.wtxn)?;
         word_pair_proximity_docids.clear(self.wtxn)?;
         word_prefix_pair_proximity_docids.clear(self.wtxn)?;
+        word_level_position_docids.clear(self.wtxn)?;
         facet_field_id_value_docids.clear(self.wtxn)?;
         field_id_docid_facet_values.clear(self.wtxn)?;
         documents.clear(self.wtxn)?;
diff --git a/milli/src/update/delete_documents.rs b/milli/src/update/delete_documents.rs
index b60b7bac2..f9303d339 100644
--- a/milli/src/update/delete_documents.rs
+++ b/milli/src/update/delete_documents.rs
@@ -330,6 +330,21 @@ impl<'t, 'u, 'i> DeleteDocuments<'t, 'u, 'i> {
 
         drop(iter);
 
+        // We delete the documents ids that are under the word level position docids.
+        let mut iter = word_level_position_docids.iter_mut(self.wtxn)?.remap_key_type::<ByteSlice>();
+        while let Some(result) = iter.next() {
+            let (bytes, mut docids) = result?;
+            let previous_len = docids.len();
+            docids.difference_with(&self.documents_ids);
+            if docids.is_empty() {
+                iter.del_current()?;
+            } else if docids.len() != previous_len {
+                iter.put_current(bytes, &docids)?;
+            }
+        }
+
+        drop(iter);
+
         Ok(self.documents_ids.len())
     }
 }